Four-year evolution of insulin regimens, glycaemic control, hypoglycaemia and body weight after starting insulin therapy in type 2 diabetes across three continents

Accepted Manuscript Title: Four-year evolution of insulin regimens, glycaemic control, hypoglycaemia and boPlease check the short title that has been created, or suggest an alternative of fewer than 80 characters including spaces.–>Four-year evolution of insulin regimens, glycaemic control, hypoglycaemia and bo–>dy weight after starting insulin therapy in type 2 diabetes across three continents Author: Philip D. Home Marie-Paule Dain Nick Freemantle Ryuzo Kawamori Martin Pfohl Sandrine Brette Val´erie Pilorget Werner A. Scherbaum Giacomo Vespasiani Maya Vincent Beverley Balkau PII: DOI: Reference:

S0168-8227(15)00050-9 http://dx.doi.org/doi:10.1016/j.diabres.2015.01.030 DIAB 6289

To appear in:

Diabetes Research and Clinical Practice

Received date: Revised date: Accepted date:

29-7-2014 24-10-2014 18-1-2015

Please cite this article as: P.D. Home, M.-P. Dain, N. Freemantle, R. Kawamori, M. Pfohl, S. Brette, V. Pilorget, W.A. Scherbaum, G. Vespasiani, M. Vincent, B. Balkau, Four-year evolution of insulin regimens, glycaemic control, hypoglycaemia and body weight after starting insulin therapy in type 2 diabetes across three continents, Diabetes Research and Clinical Practice (2015), http://dx.doi.org/10.1016/j.diabres.2015.01.030 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Four-year evolution of insulin regimens, glycaemic control, hypoglycaemia and body weight after starting insulin therapy in type

ip t

2 diabetes across three continents

cr

Philip D. Home a,*, Marie-Paule Dain b, Nick Freemantle c, Ryuzo Kawamori d, Martin Pfohl e, Sandrine Brette f, Valérie Pilorget b, Werner A. Scherbaum g,

us

Giacomo Vespasiani h, Maya Vincent b, Beverley Balkau i

Newcastle University, Newcastle upon Tyne, United Kingdom

b

Sanofi, Paris, France

c

Department of Primary Care and Population Health, University College London, United Kingdom

d

Department of Medicine, Juntendo University, Tokyo, Japan

e

Evangelisches Bethesda-Krankenhaus zu Duisburg GmbH, Duisburg, Germany

M

ed

f

an

a

Lincoln, Boulogne-Billancourt, France

Heinrich-Heine-University, University Hospital Düsseldorf, Düsseldorf, Germany

h

Diabetology and Metabolic Disorders Centre, Madonna del Soccorso Hospital, San Benedetto

i

Ac ce

del Tronto, Italy

pt

g

INSERM, U1018, University Paris Sud 11, UMRS 1018, F-94807, Villejuif, France

* Corresponding author: Professor Philip Home, Institute for Cellular Medicine-Diabetes, The Medical School, Framlington Place, Newcastle upon Tyne, NE2 4HH, United Kingdom Phone: + 44 191 208 7154/8643; Fax: + 44 191 208 0723 E-mail: [email protected]

1

Page 1 of 28

Journal: Diabetes Research & Clinical Practice Word count (5000): 5374 Tables (limit 5, combo of figs/tables): 2

ip t

Figures: 2 (plus 1 supplemental)

Ac ce

pt

ed

M

an

us

cr

Citations (50 limit): 27

2

Page 2 of 28

Abstract Aims: It is of interest to understand how insulin therapy currently evolves in clinical practice, in

cr

evolution prospectively over 4 years, to assist health care planning.

ip t

the years after starting insulin in people with type 2 diabetes. We aimed to describe this

Methods: People who had started any insulin were identified from 12 countries on three

us

continents. Baseline, then yearly follow-up, data were extracted from clinical records over 4 years.

an

Results: Of the 2999 eligible people, 2272 were followed over 4 years. When starting insulin, mean (SD) duration of diabetes was 10.6 (7.8) years, HbA1c 9.5 (2.0) % (80 [22] mmol/mol) and

M

BMI 29.3 (6.3) kg/m2. Initial insulin therapy was basal 52%, premix 23%, mealtime + basal 14%, mealtime 8% and other 3%; at 4 years, 30%, 25%, 33%, 2% and 5%, respectively, with 5% not

ed

on insulin. Insulin dose was 20.2 U/day at the start and 45.8 U/day at year 4. There were 1258 people (55%) on their original regimen at 4 years, and this percentage differed according to

pt

baseline insulin regimen. HbA1c change was –2.0 (2.2) % (–22 [24] mmol/mol) and was similar

Ac ce

by final insulin regimen. Hypoglycaemia prevalence was <20% in years 1 to 4. Body weight change was mostly in year 1, and was very variable, mean +2.7 (7.5) kg at year 4. Conclusion: Different insulin regimens were started in people with differing characteristics, and they evolved differently; insulin dose, hypoglycaemia and body weight change were diverse and largely independent of regimen.

Keywords: CREDIT, non-interventional study, insulin regimens, glucose-lowering mediations, glycaemic control, hypoglycaemia

3

Page 3 of 28

1.

Introduction

When people with type 2 diabetes on metformin and other oral therapies can no longer attain

ip t

the recommended glucose control targets, insulin therapy is generally started. Recommendations as to insulin type vary [1-3]. Some authorities endorse both basal and

premix approaches, and others place more emphasis on basal insulin; mealtime insulin alone is

cr

used on occasion, or in a comprehensive mealtime + basal regimen even from the time of

starting insulin [4]. Randomized clinical trials have provided evidence on the relative efficacy of

us

insulin regimens [5,6], but these trials were conducted in people and clinical environments that may not be representative of those in routine clinical practice. Little is known of the impact of starting insulin for outcomes of importance to people with diabetes, such as change in insulin

an

regimen, insulin dose, hypoglycaemia and weight change, with data available only for single countries ([7-9]), or in less or recently developed nations [4], or particular insulins [10,11], often

M

with follow-up of one year or less [4,7,8].

Analyses based on non-interventional observational studies may overcome some of the challenges of generalizability inherent to randomized trials and such studies provide a bridge

ed

from randomized trials towards more routine settings [4]. The Cardiovascular Risk Evaluation in people with Type 2 Diabetes on Insulin Therapy (CREDIT) study, an international 4-year, noninterventional, longitudinal study, was designed to evaluate prospectively, in routine clinical

pt

practice in a large number of clinical centres, the relationship between blood glucose control and cardiovascular events in people beginning any insulin. Further, this study aimed to describe

Ac ce

current medical practice in people with type 2 diabetes using insulin. Here we report the evolution of insulin use, associated blood glucose-related outcomes and effects on body weight and hypoglycaemia, over 4 years. Cardiovascular outcomes and the factors associated with them will be reported separately.

2.

Methods

The CREDIT study design, site/participant selection process and participant baseline characteristics have been reported previously [12]. Briefly, the study was conducted in 314 centres in 12 countries—Canada, Japan and 10 in Europe—between December 2006 and May 2012. Ethical approval was obtained for all study sites. Conduct of the study adhered to 4

Page 4 of 28

standards of data collection for clinical trials, according to the Declaration of Helsinki. Prior written informed consent was obtained from all participants. Men and women with type 2 diabetes, age >40 years, were eligible if they had started any insulin regimen >1 month and <12 months prior to study entry and had an HbA1c measurement

ip t

within 3 months of beginning insulin. Data at the start of insulin were collected retrospectively from clinical records. As CREDIT was a non-interventional study, there was no fixed study visit schedule, and insulin choice, dosage, titration, funding and concomitant oral agent therapy were

cr

according to usual local practice. Data were gathered from routine clinical practice, with the physicians asked to report updated data every 6 months. Data at ‘4 years’ represent that

us

ascertained 42–54 months after beginning insulin, and data for 1, 2 or 3 years were for 9–18, 18–30 and 30–42 months, respectively.

an

Glucose control was assessed by HbA1c, fasting plasma glucose (FPG) and postprandial plasma glucose (PPPG). HbA1c is presented in both National Glycohemoglobin Standardization Program and International Federation of Clinical Chemistry units [13,14]. FPG and PPPG are

M

reported as either laboratory or self-monitored glucose values. Documented symptomatic hypoglycaemia, nocturnal hypoglycaemia and severe hypoglycaemia were assessed over the 6 months prior to the follow-up date. Body weight change, other glucose-lowering medications

2.1

Statistical methods

ed

and changes of insulin regimen were also assessed.

pt

Analyses were performed with SAS software, version 9.1 (Cary, NC, USA). All data are reported and analysed using descriptive statistics. Data on insulin regimens, other glucose-lowering

Ac ce

medications, blood glucose control and body weight are presented by insulin regimen when starting insulin. For 4-year and change from baseline data, data for insulin dose, blood glucose control, body weight and hypoglycaemia are presented by insulin regimen at 4 years. Initial and yearly data for insulin dose, HbA1c, FPG, PPPG and body weight change are presented for the entire population. Updated HbA1c, defined as the average of all values from 1 month after starting insulin to 4 years of follow-up, was used to report results by baseline insulin regimen; all other results for HbA1c are reported by year interval. Because of the likelihood of some degree of allocation bias, comparative statistical testing was not performed, avoiding the danger of spurious statistically significant findings with the large numbers of people studied.

3.

Results 5

Page 5 of 28

There were 3060 participants included, with 2999 having valid eligibility criteria (Supplementary, Fig. S1). When starting insulin, the eligible population was 49% female, with an average (standard deviation [SD]) age of 61 (10) years, body mass index of 29.3 (6.3) kg/m 2, duration of diabetes of 11 (8) years and HbA1c of 9.5 (2.0) % (80 [22] mmol/mol). Seventy percent had at least one microvascular disease, 34% had a macrovascular disease and 69% were previously

ip t

diagnosed with hypertension. In the 4-year data interval, data were available for 2272 (75.8%) participants. Evolution of insulin regimens

cr

3.1

Of the 2999 participants starting insulin, 52.0% started on a basal insulin, 14.0% on mealtime +

us

basal, 7.5% on mealtime, 23.1% on premix and 3.4% on other insulin (premixed insulin + basal and/or mealtime insulin) (Table 1). For the 2272 participants with 4-year data, the distribution of

an

the starting insulin regimens was similar to that for all eligible participants (Fig. 1). At 4 years, 605 (52.0%) of the 1163 participants who started on basal insulin were on basal insulin, 515 (44.3%) moved to other regimens and 43 (3.7%) stopped insulin (Fig. 1). Sixty-six participants

M

on other regimens changed to basal insulin, so that in all 671 (29.5% of participants with 4-year data) were on basal insulin at 4 years (Fig. 1). For the 331 participants who started on mealtime + basal insulin, 222 (67.1%) were on mealtime + basal insulin at 4 years, 83 (25.0%) moved to

ed

other regimens and 26 (7.9%) stopped insulin. There were 519 participants who changed to mealtime + basal insulin, for a total of 741 (32.6%) on the regimen at 4 years (Fig. 1). Of the 536 participants who started on premix insulin, 362 (67.5%) were on premix at 4 years, 148

pt

(27.6%) moved to other regimens and 26 (4.9%) stopped insulin (Fig. 1). However, 210

Ac ce

participants changed to premix, so that in all, 572 (25.2%) were on premix at 4 years. Only 33 (18.2%) of the 181 participants who started on mealtime insulin were on mealtime insulin at 4 years, while 126 (69.6%) moved to other insulin regimens and 22 (12.2%) stopped insulin (Fig. 1). Twenty-one participants moved to mealtime insulin for a total of 54 (2.4%) on the regimen at 4 years. Thirty-six (59.0%) of the 61 participants who started on other insulin were on other insulin at 4 years, while 22 (36.1%) moved to other insulins and 3 (4.9%) stopped insulin. Seventy-eight participants moved to other insulin for a total of 114 (5.0%) on the regimen at 4 years. A total of 120 participants (5.3%) from all of the regimens had stopped insulin at 4 years. In all, 1258 (55.4%) of the 2272 participants with 4-year data were on the same starting regimen at 4 years.

6

Page 6 of 28

3.2

Evolution of other glucose-lowering medications

Of the eligible participants starting insulin, other glucose-lowering medication(s) were taken by 88.7% starting on basal insulin, 34.8% on mealtime + basal insulin, 44.7% on mealtime insulin, 61.8% on premix and 33.0% on other insulin (Table 1). The largest percentage taking two other

ip t

medications was for those starting a basal insulin (43%), while the largest percentage (67%) of those taking no other glucose-lowering medication was for the other insulin grouping (Table 1). When analysed by insulin regimen at 4 years, 88.4% of those on basal insulin, 46.6% on

cr

mealtime + basal insulin, 44.4% on mealtime insulin, 58.2% on premix, 46.5% on other insulin and 86.7% of those who stopped insulin were taking at least one other glucose-lowering

us

medication at 4 years (Table 2). The largest group of those on basal insulin at 4 years were those taking one other glucose-lowering medication (43%). For other regimens, the largest

an

group was people taking no other glucose-lowering medication (range 42%–56%, Table 2). Metformin and sulfonylureas were the predominant glucose-lowering medications taken when starting insulin; the frequencies of other medications ranged from 0%–12% across

M

regimens (Table 1). Metformin was taken by 63.7% of those starting on basal insulin, 25.0% starting on mealtime + basal insulin, 30.5% on mealtime insulin, 42.1% on premix and 22.3% on other insulin, while sulfonylureas were taken by 63.1%, 11.7%, 21.7%, 30.9% and 11.7%,

ed

respectively (Table 1). When analysed by insulin regimen at 4 years, 65.1% of those on basal insulin, 39.0% on mealtime + basal insulin, 33.3% on mealtime insulin, 45.6% on premix, 33.3% on other insulin and 53.3% of those who stopped insulin were taking metformin, while 45.2%,

pt

7.6%, 16.7%, 13.6%, 8.8% and 45.0%, respectively, were taking sulfonylureas (Table 2).

Ac ce

Overall for participants starting insulin, 30.3% were not taking any other glucose-lowering medications, increasing to 36.1% at 4 years. The use of sulfonylureas declined from 43.6% at the start of insulin to 22.4% at 4 years, though for those on basal insulin, 45% still used sulfonylureas. Thiazolidinedione use also declined (8.7% vs. 4.5%). Metformin use remained relatively constant (49.4% vs. 48.7%). The use of GLP-1 receptor agonists and DPP-4 inhibitors increased from ≤0.1% to 2.4% and 5.4%, respectively. 3.3

Evolution of insulin dose

For the entire population, the mean (SD) daily dose of insulin upon starting was 20.2 (14.9) U/day and increased to 45.8 (31.7) U/day at 4 years (Fig. 2). By insulin regimen at 4 years, the starting dose ranged from 15.7 (11.0) U/day for those ending on basal insulin to 23.3 (17.0) U/day for those on mealtime + basal insulin (Table 2). The dose at 4 years ranged from 27.2 7

Page 7 of 28

(41.2) U/day on mealtime insulin to 59.8 (34.5) U/day on mealtime + basal insulin. The starting dose for participants who were not taking insulin at 4 years was 17.2 (13.4) U/day. 3.4

Blood glucose control

Overall mean (SD) HbA1c declined from 9.5 [2.0] % (80 (21) mmol/mol) when starting insulin to

ip t

7.6 [1.3] % (60 (14) mmol/mol) at 4 years (median (7.3 %) [56 mmol/mol] (Fig. 2). The change in HbA1c at 4 years was –2.0 [2.2] % (–22 (24) mmol/mol). FPG declined from 11.6 (3.8) mmol/l at the start to 7.7 (2.4) mmol/l at 4 years (Fig. 2). PPPG declined from 14.3 (4.6) mmol/l to 9.4

cr

(3.2) mmol/l at 4 years (Fig. 2). The change at 4 years was –3.8 (4.0) mmol/l for FPG and –5.1 (4.9) mmol/l for PPPG. Most of the decline in HbA1c (–20 [23] mmol/mol (–1.8 [2.1] %)), FPG (–

us

3.5 [3.9] mmol/l) and PPPG (–4.6 [4.9] mmol/l) occurred in the first year (Fig. 2).

The people started on mealtime + basal or premix regimens had average HbA1c at baseline

an

rather higher than those started on basal or mealtime insulin alone (Table 1). However, over the 4 years, updated HbA1c, which averaged 59 (11) mmol/mol (7.6 [1.0] %), appeared little different according to insulin regimen at the start, basal 61 (11) mmol/mol (7.7 [1.0] %),

M

mealtime + basal 58 (12) mmol/mol (7.5 [1.1] %), mealtime only 57 (11) mmol/mol (7.4 [1.0] %) and premix 62 (12) mmol/mol (7.8 [1.1] %). Also by insulin regimen at start HbA1c to a target of <7.0 % (<53 mmol/mol) was achieved at 1 and 4 years by basal: 31% and 34%, mealtime +

pt

and 36%.

ed

basal: 42% and 42%, mealtime only: 45% and 44%, premix: 28% and 32% and no insulin 36%

By insulin regimen at 4 years, the baseline mean HbA1c ranged from 9.1% (76 mmol/mol)

Ac ce

for those ending on basal insulin to 9.8% (84 mmol/mol) for those ending on premix or no insulin (Table 2). At 4 years, mean HbA1c ranged from 7.3% (56 mmol/mol) for those on basal alone, mealtime alone or no insulin to 7.7% (61 mmol/mol) on premix or mealtime + basal insulin. By final insulin regimen, the mean change from baseline was similar for the different insulin regimens (Table 2).

Mean FPG at the start of insulin therapy ranged from 9.9 mmol/l for those on mealtime insulin at 4 years to 11.6 mmol/l on premix (Table 2). At 4 years, mean FPG ranged from 7.2 mmol/l on basal insulin to 8.4 mmol/l on mealtime insulin, with mean change from –1.1 mmol/l on mealtime insulin to –4.1 mmol/l on basal insulin (Table 2). Mean PPPG at the start of insulin therapy ranged from 13.2 mmol/l for those on basal insulin at 4 years to 15.9 mmol/l for those who stopped insulin (Table 2). At 4 years, mean PPPG ranged from 8

Page 8 of 28

8.9 mmol/l for those on basal insulin to 9.8 mmol/l on premix, with mean change from –4.4 mmol/l with basal insulin to –6.4 mmol/l for those who stopped insulin (Table 2). 3.5

Body weight change and hypoglycaemia

Body weight at 4 years increased for the entire population by 2.7 (7.5) kg, with most of the gain

ip t

occurring in the first year (1.9 [5.2] kg) (Fig. 2). Initial mean (SD) body weight ranged from 74.1 (18.5) kg for those on other insulin at 4 years to 83.3 (18.4) kg on basal insulin (Table 2). Body weight at 4 years ranged from 75.7 (22.2) kg for those on no insulin to 84.6 (18.3) kg on basal

cr

insulin. The change ranged from –2.2 (7.7) kg for those on no insulin to + 4.2 (7.6) kg on

us

mealtime + basal insulin (Table 2).

In the entire population, the percentage of participants who experienced at least one hypoglycaemia episode in the last 6 months of year 1 was 18.5%; year 2, 18.3%; year 3, 16.1%

an

and year 4, 16.6% (Fig. 2). The mean (SD) number of documented symptomatic hypoglycaemia episodes per person during the last 6 months of year 1 were 1.1 (4.6); year 2, 1.0 (4.8); year 3, 0.9 (3.2) and year 4, 1.03 (3.88), while the number of documented nocturnal symptomatic

M

hypoglycaemia episodes per person were for the last 6 months of year 1, 0.1 (0.9); year 2, 0.1 (0.9); year 3, 0.1 (0.8) and year 4, 0.21 (1.17). The proportion of participants who reported at year 3, 2.0% and year 4, 2.0%.

ed

least one severe hypoglycaemia episode in the last 6 months of year 1 was 1.9%; year 2, 1.7%;

The percentage of participants with at least one documented hypoglycaemia episode in the

pt

last 6 months before the 4-year follow-up ranged from 14.8% for those on mealtime insulin alone to 19.1% for those on premix (Table 2). The number of documented symptomatic and

Ac ce

nocturnal hypoglycaemia episodes per person was also similar across insulin regimens. The percentage of participants who reported at least one severe hypoglycaemic episode during that same period ranged from none on other insulin or no insulin to 5.6% on mealtime insulin (Table 2). The average number of events per person for severe hypoglycaemia in the last 6 months before 4 years ranged from 0.00 (0.00) on other insulin or no insulin to 0.22 (1.37) with mealtime insulin.

Even on basal insulin at 4 years, with 45% using a sulfonylurea, only 10 people experienced severe hypoglycaemia, and the rate was the same (3.3%) in sulfonylurea and non-sulfonylurea users. In basal insulin users at 4 years the proportion of people having documented symptomatic hypoglycaemia in the last 6 months was independent of sulfonylurea use or nonuse (14.9% for both), as was event rate and nocturnal event rate (data not shown). Use of 9

Page 9 of 28

sulfonylureas at 4 years in all other insulin groups was too small (except no insulin) for rates of documented symptomatic hypoglycaemia to be meaningful.

4.

Discussion

The CREDIT study, conceived both to examine the performance of different insulin regimens

ip t

over 4 years and to examine relationships of those regimens to cardiovascular outcomes (to be reported separately), offers a unique opportunity to study how insulin regimens and health

outcomes related to them, evolve in routine clinical practice. This is of importance for people

cr

with diabetes starting insulin and for the health professionals advising them, but data are also body weight change, hypoglycaemia and insulin dose.

us

provided for health care planning and health economic analysis with descriptive information on

The study was performed in nations with established economic development (Canada,

an

Europe, Japan), and there were no restrictions as to insulin types or regimens that could be used, and no protocol-determined interventions or assessments other than those performed in normal clinical practice. Other studies that have looked at people starting insulin have generally

M

been restricted to particular insulin types, such as insulin analogues, have been single centre, and/or have been shorter term, though some have addressed therapy performance in developing/recently prosperous countries [4,9-11,15]. Additionally, some longer term

ed

randomized control data are available, but the levels of intervention were unusually high throughout, with the explicit intention of influencing health outcomes [5,16].

pt

We have noted previously, in discussing factors influencing choice of first insulin in this population, that overall blood glucose control was very poor at the time of starting insulin [12]. It

Ac ce

is possible that this reflects the finding of poor glucose control in circumstances which lead to insulin being started, such as hospitalization or after specialist referral, rather than being typical of ambulatory care, but it is consistent with findings of the A1chieve study, with the UK database studies, and with US data [4,8,17,18]. While numerically the HbA1c levels before starting insulin were highest for the initial insulin regimens that combined a meal-time and a basal insulin (including premix), and while this might reflect physicians’ perception that people in poorer control required more comprehensive insulin replacement, because this is a non-randomized study the observation does not lend itself to statistical comparison [19]. Despite poor control at baseline, or perhaps because of it, response to all insulin regimens was good at 1 year and through to 4 years, with mean HbA1c reductions of 1.8–2.1 % units (20–23 mmol/mol) in those remaining on insulin, and no mean change between 1 and 4 years. 10

Page 10 of 28

The high baseline values are consistent with large initial HbA1c reductions even for non-insulin therapies [20-22]. These large initial falls in HbA1c have been reported in other observational studies, and attributed as much to educational/behavioural change as to the insulin itself [4], though in clinical database studies, poorer glucose is reported [8], and is perhaps associated with greater persistence of initial therapy [8,23,24]. The average levels achieved were not to

ip t

internationally endorsed levels: <7.0% (<53 mmol/mol), but the population studied had a long duration of diabetes, with significant comorbidities. It is likely that in line with modern guidelines,

cr

individual targets were higher in a sizeable proportion of the study participants [1,2,25].

The stability of glucose control in routine diabetes care over 4 years is impressive, given

us

that islet β-cell function deteriorates with time [26], and that insulin therapy in the UK Prospective Diabetes Study (UKPDS) showed continuing deterioration of HbA1c [27]. However, it is known that, given attention, glucose control need not deteriorate with time, even late in the

an

course of diabetes, as demonstrated in the Action to Control Cardiovascular Risk in Diabetes (ACCORD) study, for both the standard and the intensive control groups [28]. It is however clear

M

in CREDIT that both the initial improvement in control and later maintenance of control were associated with continuing increases in insulin dose. Barriers to maintenance of glucose control on insulin are often cited as weight gain and hypoglycaemia, but in our study weight gain was

ed

very modest in those continuing to be treated with mealtime or basal insulin alone, and overall was about 1 kg a year in those requiring some form of combined therapy including premix. This is in contrast to the UK study, with weight gain of around 3 kg in 12 months despite worse

pt

glucose control [8], perhaps emphasizing that lifestyle input had a greater impact in our multinational study. Predictors of body weight change in the first year of the CREDIT study

Ac ce

have been published [29].

Hypoglycaemia did not increase after the first year, continuing to affect fewer than 20% of people over the last 6 months of the study, with a population rate of approximately one episode per person. Severe hypoglycaemia (requiring assistance) was uncommon, and thus the estimates of its frequency uncertain, but our data suggest a rate of about 0.2 episodes per year on average, assuming ascertainment is satisfactory. It is not clear what proportion of this severe hypoglycaemia might require paramedical or medical assistance, glucagon, or association with an adrenergic reaction. The rate is lower than that reported by a UK study group, but the authors of that smaller study are noted for a special interest in hypoglycaemia and may have been following unusual populations of people at higher risk [30]. However, low rates of severe hypoglycaemia have been reported in populations starting insulin even in short-term 11

Page 11 of 28

observational studies [4,31], and the proportion of people in CREDIT suffering any event is consistent with the Treating To Target in Type 2 diabetes (4T) trial [5]. Some people, especially in the basal insulin only group, remained on sulfonylureas throughout. That their prevalence of hypoglycaemia in the last 6 months was identical to non-sulfonylurea users can probably be attributed to clinical care, the previous 3.5 years allowing opportunity for stopping any

ip t

medication which was a problem. This might also account for the low percentage of sulfonylurea users in the those taking premix or a meal-time+basal regimen at the end of study, or it might be

cr

that position statements advising against this were being followed [2]. Alternatively in people with reasonable and stable blood glucose control (mean HbA1c at 4 years 7.2 %), use of

us

modern sulfonylureas in combination with basal insulin might not add to the sulfonylurea risk. It is to some extent encouraging and reassuring that in routine care, clinicians are able to help people with diabetes and to use insulin to good effect. However, the standard deviations of

an

insulin dose and body weight change, as well as for hypoglycaemia, reveals that there are individuals who have more problems on insulin, and these problems probably influence general

M

clinical perceptions to some degree. Thus the standard deviation for weight change at 4 years was ±6–7.8 kg, while the wide standard deviation for 4-year insulin dose suggests more than one in 20 individuals were using >120 U/day. In the circumstance where only one in six

ed

individuals had a hypoglycaemic event in 6 months, the average rate is ~1 event, the group of affected individuals is experiencing approximately one event per month, and a small minority probably a higher rate. Thus while the overall experience of insulin therapy over 4 years is good,

pt

this is not to deny that a small proportion of people have problems. By either initial or final insulin regimen, outcomes were fairly similar, with the exception of

Ac ce

body weight, as noted above. However, only half of the people initially on basal insulin, and about two-thirds on multiple injection therapy or premix, were still on their first regimen by the end of study. As might be expected, the majority starting basal who moved to a combined regimen did so by adding mealtime insulin; a much smaller group went to a premix regimen. Those starting on basal insulin alone tended to be heavier, and avoidance of any mealtime insulin may have influenced physicians' choice of insulin, but it is also possible that confounding of the data is occurring due to concurrence of heavier populations in regions where use of basal regimens have become popular. Those who moved away from a mealtime + basal regimen went almost equally to basal alone or premix, but in this observational study we cannot know if these changes are driven by desire for regimen simplification, high rates of hypoglycaemia with mealtime insulin, or because the complex regimen was not necessary in those individuals. 12

Page 12 of 28

However, as a result, the best glucose control levels were for those remaining on single insulin regimens, with medians of 54 mmol/mol (7.1%) and 52 mmol/mol (6.9%) for basal and mealtime alone, respectively. The largest proportion moving from premix go to a mealtime + basal regimen, but again

ip t

whether this is driven by poor blood glucose control, desire for more flexible eating or a problem with hypoglycaemia is unknown. However, the observation that overall approximately 50% of people did change insulin regimen implies that physicians took an activist approach to insulin

cr

therapy once started, as noted for dose titration above. Clearly the temptation merely to adjust insulin dose was avoided, and seemingly by the diversity of changes, attempts were generally

us

made to match insulin regimens to the individuals' evolving biomedical needs and personal desires.

an

A proportion of people (~5%) came off insulin therapy altogether at some point over the 4 years. As these people had a greater fall in HbA1c than those who stayed on insulin, and a fall in body weight, it seems their cessation of insulin therapy was in response to lifestyle and

M

biomedical changes, rather than dissatisfaction with insulin. Consistent with this, the starting insulin dose of this group was not low, baseline oral agent number was high and baseline body weight was not low.

ed

Use of oral agents showed some unexpected characteristics. Metformin use was not usual in combination with insulin, and, while it increased during the study, it never exceeded two-thirds

pt

of the population. While our relatively older study population would include a significant percentage with some renal impairment, the data more likely reflect the belief of prior decades

Ac ce

that insulin therapy should be used alone. The increase in metformin use over the 4 years of study is then consistent with wider adoption of more recent guideline advice [1-3]. More easily explained is the higher use of sulfonylureas with basal insulin at the start of therapy, though also to some extent with premix, and the fall in number of users with time, particularly on the latter regimen. However, the extent to which these changes are driven by experience of hypoglycaemia or weight gain, or guidelines, cannot be known. Other glucose-lowering therapies were little used with insulin, and the small trends in thiazolidinediones and incretin therapies are merely consistent with safety concerns and license changes over the period of study. Our study has other limitations. As noted, it was restricted to economically developed nations. Furthermore, even between and within these countries, clinical habits vary markedly. This can lead to loss of generalizability, but also bias, for example, by the higher initial use of 13

Page 13 of 28

thiazolidinediones in Canada, or the greater use of premix and lower body mass index in Japan. Relying on local routine measurements would account for some of our apparent loss to followup, as the individual participants might not have had a clinical assessment in the time windows used for our yearly data collection. We also assume that laboratory measurements are comparable across countries and sites, something that has been a problem, in particular with

ip t

HbA1c [14].

Nevertheless, these data are perhaps the best available to inform further clinical guidelines

cr

about the use of insulin therapy, for planning and for economic assessments, and to guide

people with diabetes and their advisors about what to expect. We would also conclude that

us

despite insulin being started at levels of glucose control much worse than generally advised, the results overall are respectable, if not perfect, in terms of achieved metabolic control, tolerability and safety, particularly when taken in the context of the Outcome Reduction with Initial Glargine

an

Intervention (ORIGIN) study, undertaken in a very different cohort of people with diabetes [16]. Most encouraging perhaps is the evidence that clinicians in routine practice are taking adequate

M

steps to change insulin dose and regimens to avoid deterioration of blood glucose control with time, and actively changing and sometimes even stopping insulin therapy appropriately in response to other needs.

ed

In conclusion, in regular clinical practice across a group of developed nations, clinicians and people with diabetes achieve modestly encouraging results in regards of glucose control, weight change, and hypoglycaemia when starting a range of insulin regimens. Glucose control does

pt

not deteriorate over 4 years, this being attributable to evolution in insulin regimens in over half

Ac ce

the people starting insulin, as well as insulin dose titration.

14

Page 14 of 28

Acknowledgements This study was supported by Sanofi. Editorial support was provided by Tom Claus, PhD, of

ip t

PPSI (a PAREXEL company) and was funded by Sanofi. P.D.H, either personally or through institutions with which he is associated, receives funding for research, advisory and educational activities from most insulin and oral-agent manufacturers,

cr

including Sanofi. N.F. has received research grants and served as consultant to Eli Lilly,

Medtronic, Novo Nordisk, Pfizer and Sanofi, and has served on speaker bureaus for Novo

us

Nordisk and Sanofi. R.K. has received honoraria for lectures from Eli Lilly, Novo Nordisk and Sanofi. M.P. has served on advisory boards for Boehringer Ingelheim, Eli Lilly, Novo Nordisk

an

and Sanofi, on speaker bureaus for Novo Nordisk and Sanofi, and received research grants from Eli Lilly, Medtronic, Novo Nordisk, Pfizer and Sanofi. W.A.S. has no conflicts to declare. G.V. has served on advisory boards for Sanofi, speaker bureaus for Lifescan and Novo Nordisk,

M

and as consultant for Meteda. B.B. has served on advisory bureaus for Boehringer Ingelheim, Bristol-Myers Squibb, Eli Lilly and Sanofi, and received research grant from Servier. M.-P.D.,

ed

V.P. and M.V. are employees of Sanofi. S.B. is a consultant to Sanofi. P.D.H. contributed to study design, study conduct/data collection, data analysis and writing. N.F. contributed to the study design, study conduct/data collection, data analysis and writing. B.B.

pt

contributed to the study design, conduct, and data collection and critically re-reading the manuscript. W.A.S. contributed to the study design, study conduct/data collection, data analysis

Ac ce

and writing. M.V. contributed to data analysis and writing. M.P. contributed to the study design and writing. G.V. contributed to the study design, data collection and writing. R.K. contributed to study design, study conduct/data collection, data analysis and writing. M.-P.D. contributed to study design, study conduct/data collection, data analysis and writing. S.B. contributed to data analysis and writing. V.P. contributed to study design, study conduct/data collection, data analysis and writing.

Parts of this study were presented at the 49th Annual Meeting of the European Association for the Study of Diabetes, Barcelona, Spain, 23–27 September 2013, the International Diabetes Federation’s World Diabetes Congress, Melbourne, Australia, 02–06 December, 2013, the American Diabetes Association Annual Scientific Meeting, San Francisco, 13–17 June 2014, and the European Association for the Study of Diabetes, Vienna, Austria, 16–19 September 2014.

15

Page 15 of 28

References

ip t

1. International Diabetes Federation: Global Guideline for Type 2 Diabetes. Diabetes Res Clin Pract. 2014. 104 (1): 1-52

us

cr

2. Inzucchi SE, Bergenstal RM, Buse JB, Diamant M, Ferrannini E, Nauck M, et al. Management of Hyperglycemia in Type 2 Diabetes: A Patient-Centered Approach: Position Statement of the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD). Diabetes Care 2012;35(6):1364-79.

an

3. NICE. Blood-glucose-lowering therapy for type 2 diabetes. National Institute for Health and Care Excellence. http://pathways.nice.org.uk/pathways/diabetes#path=view%3A/pathways/diabetes/bloodglucose-lowering-therapy-for-type-2-diabetes.xml&content=view-node%3Anodes-startinginsulin-particularly-if-hyperglycaemia-is-marked. Accessed June 12, 2013.

M

4. Home P, Naggar NE, Khamseh M, Gonzalez-Galvez G, Shen C, Chakkarwar P, et al. An observational non-interventional study of people with diabetes beginning or changed to insulin analogue therapy in non-Western countries: the A1chieve study. Diabetes Res Clin Pract 2011;94(3):352-63.

ed

5. Holman RR, Thorne KI, Farmer AJ, Davies MJ, Keenan JF, Paul S, et al. Addition of biphasic, prandial, or basal insulin to oral therapy in type 2 diabetes. N Engl J Med 2007;357(17):1716-30.

Ac ce

pt

6. Strojek K, Bebakar WM, Khutsoane DT, Pesic M, Smahelova A, Thomsen HF, et al. Oncedaily initiation with biphasic insulin aspart 30 versus insulin glargine in patients with type 2 diabetes inadequately controlled with oral drugs: an open-label, multinational RCT. Curr Med Res Opin 2009;25(12):2887-94. 7. Eliasson B, Ekstrom N, Bruce WS, Oden A, Fard MP, Svensson AM. Metabolic effects of Basal or premixed insulin treatment in 5077 insulin-naive type 2 diabetes patients: registrybased observational study in clinical practice. Diabetes Ther 2014;5(1):243-54. 8. Gordon J, Pockett RD, Tetlow AP, McEwan P, Home PD. A comparison of intermediate and long-acting insulins in people with type 2 diabetes starting insulin: an observational database study. Int J Clin Pract 2010;64(12):1609-18. 9. Hall GC, McMahon AD, Dain MP, Wang E, Home PD. Primary-care observational database study of the efficacy of GLP-1 receptor agonists and insulin in the UK. Diabet Med 2013;30(6):681-6. 10. Khunti K, Damci T, Meneghini L, Pan CY, Yale JF. Study of Once Daily Levemir (SOLVE): insights into the timing of insulin initiation in people with poorly controlled type 2 diabetes in routine clinical practice. Diabetes Obes Metab 2012;14(7):654-61. 16

Page 16 of 28

11. Meneghini LF, Dornhorst A, Sreenan S. Once-daily insulin detemir in a cohort of insulinnaive patients with type 2 diabetes: a sub-analysis from the PREDICTIVE study. Curr Med Res Opin 2009;25(4):1029-35.

ip t

12. Freemantle N, Balkau B, Danchin N, Wang E, Marre M, Vespasiani G, et al. Factors influencing initial choice of insulin therapy in a large international non-interventional study of people with type 2 diabetes. Diabetes Obes Metab 2012;14(10):901-9.

cr

13. Jeppsson JO, Kobold U, Barr J, Finke A, Hoelzel W, Hoshino T, et al. Approved IFCC reference method for the measurement of HbA1c in human blood. Clin Chem Lab Med 2002;40(1):78-89.

us

14. Little RR, Rohlfing CL, Wiedmeyer HM, Myers GL, Sacks DB, Goldstein DE. The national glycohemoglobin standardization program: a five-year progress report. Clin Chem 2001;47(11):1985-92.

an

15. Hall GC, McMahon AD, Carroll D, Home PD. Macrovascular and microvascular outcomes after beginning of insulin versus additional oral glucose-lowering therapy in people with type 2 diabetes: an observational study. Pharmacoepidemiol Drug Saf 2012;21(3):305-13.

M

16

ed

. Gerstein HC, Bosch J, Dagenais GR, Diaz R, Jung H, Maggioni AP, et al. Basal insulin and cardiovascular and other outcomes in dysglycemia. N Engl J Med 2012;367(4):319-28.

pt

17. Hall GC, McMahon AD, Dain MP, Home PD. A comparison of duration of first prescribed insulin therapy in uncontrolled type 2 diabetes. Diabetes Res Clin Pract 2011;94(3):442-8.

Ac ce

18. Karter AJ, Moffet HH, Liu J, Parker MM, Ahmed AT, Go AS, et al. Glycemic response to newly initiated diabetes therapies. Am J Manag Care 2007;13(11):598-606. 19. Fonseca V, Davidson J, Home P, Snyder J, Jellinger P, Dyhr TA, et al. Starting insulin therapy with basal insulin analog or premix insulin analog in T2DM: a pooled analysis of treat-to-target trials. Curr Med Res Opin 2010;26(7):1621-8. 20. Deacon CF, Mannucci E, Ahren B. Glycaemic efficacy of glucagon-like peptide-1 receptor agonists and dipeptidyl peptidase-4 inhibitors as add-on therapy to metformin in subjects with type 2 diabetes-a review and meta analysis. Diabetes Obes Metab 2012;14(8):762-7. 21. DeFronzo RA, Stonehouse AH, Han J, Wintle ME. Relationship of baseline HbA1c and efficacy of current glucose-lowering therapies: a meta-analysis of randomized clinical trials. Diabet Med 2010;27(3):309-17. 22. Riddle MC, Rosenstock J, Gerich J. The treat-to-target trial: randomized addition of glargine or human NPH insulin to oral therapy of type 2 diabetic patients. Diabetes Care 2003;26(11):3080-6.

17

Page 17 of 28

23. Pfohl M, Dippel FW, Kostev K, Fuchs S, Kotowa W. Different persistence on initial basal supported oral therapy in Type 2 diabetics is associated with unequal distributions of insulin treatment regimens under real-life conditions in Germany. Int J Clin Pharmacol Ther 2010;48(11):761-6.

ip t

24. Quinzler R, Ude M, Franzmann A, Feldt S, Schussel K, Leuner K, et al. Treatment duration (persistence) of basal insulin supported oral therapy (BOT) in Type-2 diabetic patients: comparison of insulin glargine with NPH insulin. Int J Clin Pharmacol Ther 2012;50(1):24-32.

cr

25. Raz I. Guideline approach to therapy in patients with newly diagnosed type 2 diabetes. Diabetes Care 2013;36 Suppl 2:S139-S144.

us

26. United Kingdom Prospective Diabetes Study (UKPDS). 13: Relative efficacy of randomly allocated diet, sulphonylurea, insulin, or metformin in patients with newly diagnosed noninsulin dependent diabetes followed for three years. BMJ 1995;310(6972):83-8.

an

27. Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). UK Prospective Diabetes Study (UKPDS) Group. Lancet 1998;352(9131):837-53.

M

28. Gerstein HC, Miller ME, Byington RP, Goff DC, Jr., Bigger JT, Buse JB, et al. Effects of intensive glucose lowering in type 2 diabetes. N Engl J Med 2008;358(24):2545-59. 29. Balkau B, Home PD, Vincent M, Marre M, Freemantle N. Factors associated with weight gain in people with type 2 diabetes starting on insulin. Diabetes Care 2014;37(8):2108-13.

ed

30. UK Hypoglycaemia Study Group. Risk of hypoglycaemia in types 1 and 2 diabetes: effects of treatment modalities and their duration. Diabetologia 2007;50(6):1140-7.

Ac ce

pt

31. Valensi P, Benroubi M, Borzi V, Gumprecht J, Kawamori R, Shaban J, et al. Initiating insulin therapy with, or switching existing insulin therapy to, biphasic insulin aspart 30/70 (NovoMix 30) in routine care: safety and effectiveness in patients with type 2 diabetes in the IMPROVE observational study. Int J Clin Pract 2009;63(3):522-31.

18

Page 18 of 28

Table 1 – Glucose-lowering medications, glycaemic control and body weight when starting insulin by starting insulin regimen.

Mealtime

Basal

Mealtime

1558 (52)

103 (3.4)

None

176 (11.3)

274 (65.2)

125 (55.3)

264 (38.2)

69 (67.0)

≥1

1382 (88.7)

146 (34.8)

101 (44.7)

428 (61.8)

34 (33.0)

583 (37.4)

107 (25.5)

44 (19.5)

241 (34.8)

23 (22.3)

665 (42.7)

37 (8.8)

48 (21.2)

134 (19.4)

10 (9.7)

134 (8.6)

2 (0.5)

9 (4.0)

53 (7.7)

1 (1.0)

Metformin

993 (63.7)

105 (25.0)

69 (30.5)

291 (42.1)

23 (22.3)

Sulfonylurea

983 (63.1)

49 (11.7)

49 (21.7)

214 (30.9)

12 (11.7)

Thiazolidinedione

145 (9.3)

13 (3.0)

17 (7.5)

77 (11.1)

8 (7.8)

ed

M

n (%)

pt

1

≥3

Ac ce

2

226 (7.5)

692 (23.1)

an

Other glucose-lowering drugs,

420 (14.0)

Other

us

N = 2,999; n (%)

Premix

cr

+ basal

ip t

Starting insulin regimen

19

Page 19 of 28

119 (7.6)

3 (0.7)

10 (4.4)

27 (3.9)

1 (1.0)

Alpha-glucosidase inhibitor

69 (4.4)

17 (4.0)

22 (9.7)

60 (8.7)

2 (1.9)

GLP-1 receptor agonist

0 (0.0)

0 (0.0)

0 (0.0)

0 (0.0)

0 (0.0)

DPP-4 inhibitor

2 (0.1)

0 (0.0)

0 (0.0)

9.2 (1.8)

10.1 (2.1)

9.5 (2.0)

9.8 (2.0)

9.2 (1.9)

77 (20)

87 (23)

us

84 (22)

77 (21)

201 (56)

227 (68)

180 (64)

211 (69)

206 (63)

mmol/l

11.2 (3.1)

12.6 (3.8)

10.0 (3.6)

11.7 (3.8)

11.4 (3.5)

PPPG, mg/dl

239 (69)

280 (80)

276 (102)

266 (78)

257 (79)

13.3 (3.8)

15.6 (4.5)

15.3 (5.7)

14.8 (4.3)

14.3 (4.4)

mmol/mol

ed 83.1 (17.8)

cr

0 (0.0)

0 (0.0)

75.2 (18.5) 69.2 (17.3) 78.2 (19.1) 73.4 (14.8)

Ac ce

Body weight, kg

pt

mmol/l

M

FPG, mg/dl

80 (22)

an

HbA1c, % unit

ip t

Glinides

Mean (SD) or n (%).

DPP-4, dipeptidylpeptidase-4; FPG, fasting plasma glucose; GLP-1, glucagon-like peptide-1; PPPG, postprandial plasma glucose; SD, standard deviation.

20

Page 20 of 28

Table 2 – Glucose-lowering medications, insulin dose, glycaemic control, body weight and hypoglycaemia at 4 years, by insulin regimen at 4 years.

Mealtime

Mealtime

None

78 (11.6)

396 (53.4)

≥1

593 (88.4)

No insulin

572 (25.2)

114 (5.0)

120 (5.3)

M

741 (32.6)

Other

30 (55.6)

239 (41.8)

61 (53.5)

16 (13.3)

345 (46.6)

24 (44.4)

333 (58.2)

53 (46.5)

104 (86.7)

278 (37.5)

16 (29.6)

231 (40.4)

36 (31.6)

34 (28.3)

pt

671 (29.5)

us

+ basal

N = 2272; n (%)

Premix

cr

Basal

ip t

Insulin regimen at 4 years

250 (37.3)

56 (7.6)

5 (9.3)

79 (13.8)

16 (14.0)

35 (29.2)

53 (7.9)

11 (1.5)

3 (5.6)

23 (4.0)

1 (0.9)

35 (29.2)

Metformin

437 (65.1)

289 (39.0)

18 (33.3)

261 (45.6)

38 (33.3)

64 (53.3)

Sulfonylurea

303 (45.2)

56 (7.6)

9 (16.7)

78 (13.6)

10 (8.8)

54 (45.0)

Thiazolidinedione

29 (4.3)

20 (2.7)

2 (3.7)

29 (5.1)

7 (6.1)

15 (12.5)

Glinides

85 (12.7)

13 (1.8)

3 (5.6)

25 (4.4)

5 (4.4)

15 (12.5)

an

Other glucose-lowering drugs, n

ed

(%)

1

290 (43.2)

≥3

Ac ce

2

54 (2.4)

21

Page 21 of 28

26 (3.9)

23 (3.1)

2 (3.7)

49 (8.6)

8 (7.0)

10 (8.3)

GLP-1 receptor agonist

19 (2.8)

5 (0.7)

0 (0.0)

2 (0.3)

0 (0.0)

29 (24.2)

DPP-4 inhibitor

58 (8.6)

18 (2.4)

1 (1.9)

17 (3.0)

3 (2.6)

26 (21.7)

ip t

Alpha-glucosidase inhibitor

cr

Insulin dose

15.7 (11.0) 23.3 (17.0) 18.4 (16.3) 20.9 (14.1) 22.8 (17.9) 17.2 (13.4)

4 year, U/day

32.6 (22.2) 59.8 (34.5) 27.2 (41.2) 44.7 (29.6) 46.2 (27.6)

us

Start, U/day

–

0.2 (0.1)

0.3 (0.2)

0.3 (0.2)

0.3 (0.2)

0.3 (0.2)

0.2 (0.1)

4 year, U/kg/day

0.4 (0.2)

0.7 (0.3)

0.3 (0.3)

0.5 (0.3)

0.6 (0.2)

–

M

an

Start, U/kg/day

ed

HbA1c, % units/mmol/mol

Ac ce

4-year

Change

9.2 (1.8)/

9.6 (1.9)/

9.1 (1.7)/

9.8 (2.0)/

9.7 (2.1)/

9.8 (1.9)/

77 (20)

81 (21)

76 (13)

84 (22)

83 (23)

84 (21)

7.3 (1.1)/

7.7 (1.3)/

7.3 (1.3)/

7.7 (1.3)

7.6 (1.3)/

7.3 (1.4)/

56 (12)

61 (14)

56 (14)

61 (14)

60 (14)

56 (15)

–1.9 (2.0)/

–1.9 (2.2)/

–1.8 (2.2)/

–2.1 (2.2)/

–2.1 (2.4)/

–2.5 (2.3)/

–20 (22)

–20 (24)

–20 (24)

–23 (24)

–23 (26)

–27 (25)

203 (58)/

208 (64)/

178 (60)/

209 (63)/

208 (67)/

207 (69)/

11.3 (3.2)

11.5 (3.5)

9.9 (3.3)

11.6 (3.5)

11.6 (3.7)

11.5 (3.8)

pt

Start

FPG, mg/dl/mmol/l

Start

22

Page 22 of 28

142 (44)/

152 (52)/

140 (44)/

147 (43)/

139 (65)/

7.2 (2.1)

7.9 (2.4)

8.5 (2.9)

7.8 (2.5)

8.2 (2.4)

7.7 (3.6)

–73 (64)/

–66 (73)/

–20 (74)/

–69 (72)/

–63 (75)/

–65 (89)/

–4.1 (3.6)

–3.7 (4.1)

–1.1 (4.1)

–3.8 (4.0)

–3.5 (4.2)

–3.6 (4.9)

238 (71)/

258 (76)/

245 (98)/

13.2 (3.9)

14.4 (4.2)

13.6 (5.4)

161 (48)/

170 (55)/

8.9 (2.7)

9.4 (3.1)

ip t

Change

130 (38)/

cr

4 year

PPPG, mg/dl/mmol/l

Change

–79 (78)/

15.1 (4.6)

14.7 (4.6)

15.9 (4.9)

169 (50)/

176 (68)/

163 (56)/

172 (63)/

9.4 (2.8)

9.8 (3.8)

9.0 (3.1)

9.6 (3.5)

–92 (88)/

–85 (103)/

–98 (93)/

–5.1 (4.9)

–4.7 (5.7)

–5.4 (5.1)

us

285 (88)/

–102 (104)/ –115 (100)/ –5.7 (5.8)

–6.4 (5.6)

ed

–4.4 (4.3)

264 (83)/

an

4 year

271 (83)/

M

Start

pt

Body weight, kg

83.3 (18.4) 79.9 (17.3) 75.0 (24.8) 76.3 (18.2) 74.1 (18.5) 75.4 (21.9)

Ac ce

Start

4 year

Change

84.6 (18.3) 84.0 (17.8) 76.9 (25.9) 79.8 (18.7) 77.0 (17.9) 75.7 (22.2)

1.1 (7.8)

4.2 (7.6)

0.6 (7.6)

3.4 (6.5)

3.4 (6.0)

–2.2 (7.7)

138 (18.6)

8 (14.8)

109 (19.1)

18 (15.8)

2 (2.5)

Hypoglycaemia last 6 months – any documented

People with ≥1 event, n (%)

100 (14.9)

Any event rate (events/person) 1.22 (4.67) 1.03 (3.62) 0.94 (4.25) 1.02 (3.57) 0.78 (2.21) 0.33 (2.95)

23

Page 23 of 28

Nocturnal rate (events/person)

0.23 (1.15) 0.25 (1.27) 0.28 (2.06) 0.18 (1.11) 0.21 (0.92) 0.00 (0.00)

Hypoglycaemia last 6 months – severe

22 (3.3)

Severe event rate

19 (2.6)

3 (5.6)

0.15 (1.02) 0.09 (0.91) 0.22(1.37)

0 (0.0)

0 (0.0)

0.0 (0.06)

0.00 (0.00) 0.00 (0.00)

cr

(events/person)

2 (0.3)

ip t

People with ≥1 event, n (%)

0.03 (0.27) 0.02 (0.21) 0.17 (1.09) 0.00 (0.00) 0.00 (0.00) 0.00 (0.00)

us

Nocturnal rate (events/person)

an

Values are n (%) or mean (SD).

DPP-4, dipeptidylpeptidase-4; FPG, fasting plasma glucose; GLP-1, glucagon-like peptide-1; PPPG,

Ac ce

pt

ed

M

postprandial plasma glucose; SD, standard deviation.

24

Page 24 of 28

*Highlights (for review)

Home, et al. Highlights 1. We describe how insulin therapy evolves in clinical practice over 4 years in >2000 people with T2DM.

3. At 4 years this was 30%, 25%, 33%, 2% and 5%, with 5% not on insulin.

ip t

2. Initial insulin was basal 52%, premix 23%, meal-time+basal 14%, meal-time 8%, other 3%.

cr

4. HbA1c change was –2.0 (SD 2.2) %, similar by final insulin regimen, and stable over 4 years.

Ac ce p

te

d

M

an

us

5. Hypoglycaemia prevalence was <20% of people in all years (1 to 4 separately).

Page 25 of 28

Figure legend(s)

Fig. 1 – Change in insulin regimen over 4 years for those with data at baseline and 4 years (N = 2272). Horizontal arrows represent the movement of participants from one regimen to another at 4 years, with the number and percentage of participants moving from the initial regimen shown

ip t

above the arrow. Vertical arrows denote the movement of participants on a regimen when starting insulin to the same regimen at 4 years. The number and percentage shown intersecting

cr

the vertical arrows are participants who were on the same regimen at the start and at 4 years. Fig. 2 – Evolution of HbA1c (A), fasting plasma glucose (B), postprandial plasma glucose (C),

us

insulin dose (D), change in body weight (baseline 79.6 (18.9) kg) (E) over 4 years, with

an

proportion of people having at least one documented hypoglycaemia episode in the 6 months

Ac ce p

te

d

M

before each year interval (F). Data (A–E) are mean (standard deviation).

Page 26 of 28

Ac

ce

pt

ed

M

an

us

cr

i

Figure 1

Page 27 of 28

Ac ce p

te

d

M

an

us

cr

ip t

Figure 2

Page 28 of 28